U.S. patent number 10,613,063 [Application Number 15/793,388] was granted by the patent office on 2020-04-07 for methods and systems for chromatography data analysis.
This patent grant is currently assigned to Regeneron Pharmaceuticals, Inc.. The grantee listed for this patent is REGENERON PHARMACEUTICALS, INC.. Invention is credited to Hanne Bak, Scott Carver, Nathan L. Mao, John Mattila, Stefanie McDermott, Bernhard Schilling, Eric Shierly.
![](/patent/grant/10613063/US10613063-20200407-D00000.png)
![](/patent/grant/10613063/US10613063-20200407-D00001.png)
![](/patent/grant/10613063/US10613063-20200407-D00002.png)
![](/patent/grant/10613063/US10613063-20200407-D00003.png)
![](/patent/grant/10613063/US10613063-20200407-D00004.png)
![](/patent/grant/10613063/US10613063-20200407-D00005.png)
![](/patent/grant/10613063/US10613063-20200407-D00006.png)
![](/patent/grant/10613063/US10613063-20200407-D00007.png)
![](/patent/grant/10613063/US10613063-20200407-D00008.png)
![](/patent/grant/10613063/US10613063-20200407-D00009.png)
![](/patent/grant/10613063/US10613063-20200407-D00010.png)
View All Diagrams
United States Patent |
10,613,063 |
Mao , et al. |
April 7, 2020 |
Methods and systems for chromatography data analysis
Abstract
Embodiments of the present disclosure are directed to methods
and systems for assessing integrity of chromatography columns,
systems, and processes. The methods and systems can comprise one or
more of extracting a block and signal combination for analysis,
performing a transition analysis, performing one or more
statistical process controls, and/or implementing in-process
controls based on the statistical process controls.
Inventors: |
Mao; Nathan L. (Cohoes, NY),
Shierly; Eric (Castleton-on-Hudson, NY), Schilling;
Bernhard (Hudson, NY), Carver; Scott (Wynantskill,
NY), McDermott; Stefanie (Sleepy Hollow, NY), Mattila;
John (Nyack, NY), Bak; Hanne (New York, NY) |
Applicant: |
Name |
City |
State |
Country |
Type |
REGENERON PHARMACEUTICALS, INC. |
Tarrytown |
NY |
US |
|
|
Assignee: |
Regeneron Pharmaceuticals, Inc.
(Tarrytown, NY)
|
Family
ID: |
60269959 |
Appl.
No.: |
15/793,388 |
Filed: |
October 25, 2017 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20180113101 A1 |
Apr 26, 2018 |
|
Related U.S. Patent Documents
|
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
Issue Date |
|
|
62412563 |
Oct 25, 2016 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G01N
30/88 (20130101); G01N 30/8617 (20130101); G01N
30/8693 (20130101); G01N 2030/889 (20130101) |
Current International
Class: |
G01N
30/88 (20060101); G01N 30/86 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
|
|
|
|
|
|
|
WO 2009/094203 |
|
Jul 2009 |
|
WO |
|
WO 2010/019814 |
|
Feb 2010 |
|
WO |
|
Other References
Christopher Bork et al.: "Online integrity monitoring in the
Protein A step of mAb Production Processing--increasing reliability
and process robustness", Biotechnology Progress., vol. 30, No. 2,
Jan. 13, 2014, pp. 383-390. cited by applicant .
International Search Report for International Application No.
PCT/US2017/058190 dated Jan. 25, 2018 (6 pages). cited by applicant
.
Larson et al., Use of Process Data To Assess Chromatographic
Performance in Production-Scale Protein Purification Columns,
Biotechnol. Prog., 2003, 19, 485-492. cited by applicant.
|
Primary Examiner: Bui; Bryan
Attorney, Agent or Firm: Bookoff McAndrews, PLLC
Parent Case Text
CROSS-REFERENCE TO RELATED APPLICATIONS
This application claims priority to U.S. Application No. 62/412,563
filed on Oct. 25, 2016, the entire disclosure of which is herein
incorporated by reference.
Claims
What is claimed is:
1. A process control method, comprising: receiving raw
chromatography data including a plurality of signals, wherein each
signal of the plurality of signals is associated with one of a
plurality of blocks; obtaining a subset of data by selecting a
combination of a first block and a first signal from the raw
chromatography data; generating processed chromatography data by
applying a noise reduction technique to the subset of data, wherein
applying the noise reduction technique includes: selecting a
portion of the subset of data to analyze using predetermined set
points; normalizing the portion to prevent magnitude bias; using at
least one smoothing filter on the portion to generate smoothed
data; and analyzing the portion for dynamic signal errors;
generating transition data by performing a transition analysis on
the processed chromatography data, wherein performing the
transition analysis includes: generating a curve using the
processed chromatography data; and analyzing the curve to generate
performance parameters; and performing an action based on the
transition data, wherein performing the action includes generating
a notification of an event, generating an evaluation of the event,
or generating a deviation notification form.
2. The method of claim 1, wherein the raw chromatography data is
received from a chromatography process skid.
3. The method of claim 1, further comprising: performing a
chromatography column run, wherein the raw chromatography data is
received from the chromatography column run.
4. The method of claim 1, wherein each block of the plurality of
blocks corresponds to a step in a chromatography process.
5. The method of claim 1, wherein the selected combination includes
the first block, the first signal, and a second signal of the
plurality of signals.
6. The method of claim 1, further comprising: selecting the
combination of the first block and the first signal according to a
profile defining a plurality of selection criteria.
7. The method of claim 6, wherein the plurality of selection
criteria comprises: whether blocks occur at regular chromatography
cycle intervals; an extent to which one of the plurality of signals
saturates a detector; an extent to which the plurality of signals
approaches a stationary phase at a distinct level; a magnitude of
variation in the plurality of signals; or a number of inflection
points shown by the plurality of signals during a transition
phase.
8. The method of claim 1, wherein selecting the combination of the
first block and the first signal comprises selecting a primary
block and signal combination, and further comprising selecting a
secondary block and signal combination.
9. The method of claim 1, further comprising: selecting smoothed
data matching a feature of a chromatogram transition, wherein the
feature includes one of: derivative duration; maximum intensity;
duration from initiation; or expected background sensor noise.
10. The method of claim 1, further comprising: generating an
Individual chart, a Moving Range chart, or a Range chart based on
the transition data; and generating performance data by applying a
statistical process control to the Individual chart, Moving Range
chart, or Range chart, wherein performing the action based on the
transition data includes performing the action based on the
performance data.
11. The method of claim 10, wherein applying a statistical process
control to the Individual chart, Moving Range chart, or Range chart
comprises performing one of a multivariate data analysis or a
principal component analysis.
12. A chromatography method, the method comprising: performing the
method of claim 1 while running a chromatography column.
13. A process control method, comprising: receiving a selection of
raw chromatography data; generating smoothed data by applying a
noise reduction technique to the selection of raw chromatography
data, wherein the noise reduction technique comprises: selecting a
portion of the smoothed data to analyze using predetermined set
points; normalizing the portion of data to prevent magnitude bias;
using at least one smoothing filter on the portion of data to
generate smoothed data; and analyzing the portion of data for
dynamic signal errors; generating processed chromatography data by
selecting smoothed data matching a feature of a chromatogram
transition, wherein the feature of the chromatogram transition
includes: a derivative duration; a maximum intensity; a duration
from initiation; or expected background noise; and performing an
action based on the processed chromatography data, wherein
performing the action includes: generating a notification of an
event; generating an evaluation of the event; or generating a
deviation notification form.
14. The method of claim 13, wherein receiving the selection of raw
chromatography data comprises: receiving raw chromatography data
including a plurality of signals and a plurality of blocks, wherein
each signal of the plurality of signals is associated with a block;
and selecting a combination of a first block and a first signal
from the raw chromatography data.
15. The method of claim 13, further comprising: using the processed
chromatography data to generate one of an Individual chart, a
Moving Range chart, or a Range chart; and generating performance
data by applying a statistical process control to the Individual
chart, Moving Range chart, or Range chart by: performing a
multivariate data analysis; or performing a principal component
analysis, wherein performing the action based on the processed
chromatography data includes performing the action based on the
performance data.
16. A process control method, comprising: receiving raw
chromatography data including a plurality of signals, wherein each
signal of the plurality of signals is associated with one of a
plurality of blocks; obtaining a subset of data by selecting a
combination of a first block and a first signal from the raw
chromatography data; generating processed chromatography data by
applying a noise reduction technique to the subset of data, wherein
applying the noise reduction technique includes: selecting a
portion of the subset of data to analyze using predetermined set
points; normalizing the portion to prevent magnitude bias; using at
least one smoothing filter on the portion to generate smoothed
data; and analyzing the portion for dynamic signal errors;
generating transition data representative of a column integrity by
performing a transition analysis, wherein performing a transition
analysis includes: generating performance parameters, the
performance parameters including a maximum rate of change; and
based on the performance parameters, generating the transition
data; performing an action based on the transition data, wherein
performing the action includes generating a notification of an
event, generating an evaluation of the event, or generating a
deviation notification form.
17. The method of claim 16, further comprising: generating an
Individual chart, a Moving Range chart, or a Range chart based on
the transition data; and generating performance data by applying a
statistical process control to the Individual chart, Moving Range
chart, or Range chart, and wherein performing the action based on
the transition data includes performing the action based on the
performance data.
18. The method of claim 16, further comprising: selecting the
combination of the first block and the first signal according to a
profile defining a plurality of selection criteria, wherein the
plurality of selection criteria comprises: whether blocks occur at
regular chromatography cycle intervals; an extent to which one of
the plurality of signals saturates a detector; an extent to which
the plurality of signals approaches a stationary phase at a
distinct level; a magnitude of variation in the plurality of
signals; or a number of inflection points shown by the plurality of
signals during a transition phase.
19. The method of claim 16, further comprising: generating an
Individual chart, a Moving Range chart, or a Range chart based on
the transition data; and generating performance data by applying a
statistical process control to the Individual chart, Moving Range
chart, or Range chart, wherein applying a statistical process
control to the Individual chart, Moving Range chart, or Range chart
comprises performing one of a multivariate data analysis or a
principal component analysis.
20. The method of claim 16, further comprising: performing a
chromatography column run, wherein the raw chromatography data is
received from the chromatography column run, a chromatography
process skid, or both.
21. A chromatography method, the method comprising: performing the
method of claim 16 while running a chromatography column.
Description
TECHNICAL FIELD
Aspects of the present disclosure relate generally to
chromatography systems and methods, and, specifically, to
embodiments of methods and systems for chromatography data
analysis, e.g., for in-process monitoring and control of
chromatography systems.
BACKGROUND
Packed bed chromatography processes play an important role in the
production of biologic drug products. Many active biologics, such
as proteins, are purified for use in drug products using packed bed
chromatography. Chromatography column operation therefore may have
a significant effect on manufacturing critical process parameters
(CPP) and critical quality attributes (CQA). Moreover, the
complexity and size of biologics, as compared to, e.g., small
molecules, can make analyzing biologic quality and purity
relatively more difficult. Thus, monitoring the quality,
consistency, and integrity of chromatography processes and
equipment via in-process controls is important to ensure that
product quality meets any applicable standards (e.g., government
regulations).
Generally, column integrity can be determined by the uniform plug
flow of a mobile phase through a column's stationary phase (e.g.,
resin). Examples of loss of column integrity can include, for
example, evidence of channeling, headspace, fouled areas of flow,
and the like. Channeling may result when, among other things, a
mobile phase is able to travel some distance from a column inlet
towards the column's outlet without contacting the stationary
phase. Headspace may refer to, among other things, when a lateral
zone is created in a column that allows for non-plug flow of the
mobile phase. Fouled areas of flow may include dirt or other
residue on inlet or outlet frit surfaces, or on resin pores.
Several techniques exist for monitoring chromatography column
performance and integrity. Some techniques, such as the pulse
injection method for measuring height equivalent of a theoretical
plate (HETP), require buffer solutions needing special preparation.
Pulse injection techniques generally require operation of
chromatography equipment and the column outside of normal
processes, resulting in increased process time and labor. Other
techniques include monitoring critical parameters (e.g., step
yield, pre-pool volume, and maximum optical density during load) as
a part of routine production. However, setting alarm limits on
these parameters is difficult and imprecise, and may result in
false alarms or overly broad limits.
There exists a need for methods, systems, and processes for
measuring and managing column performance and integrity with
accuracy and precision, and with minimal disruption to processes.
Moreover, because of inherent differences between chromatography
columns, chromatography column cycles, and/or production lots for
any given product undergoing chromatography, there exists a need
for methods, systems, and processes with which to customize
analyses of column performance and integrity for a particular
column or columns, a particular cycle or cycles, and/or a
particular lot or lots of a product. Finally, there exists a need
for precise in-process controls that use such analyses, and for
methods and systems for responding to deviations from such
controls, so that issues with column integrity and performance may
be identified and corrected early, with minimal waste and
expense.
SUMMARY
Embodiments of the present disclosure may be directed to a process
control method, the method including: receiving raw chromatography
data including a plurality of signals, wherein each signal of the
plurality of signals is associated with one of a plurality of
blocks; obtaining a subset of data by selecting a combination of a
first block and a first signal from the raw chromatography data;
generating processed chromatography data by applying a noise
reduction technique to the subset of data; generating transition
data by performing a transition analysis on the processed
chromatography data; and performing an action based on the
transition data.
In some embodiments, the method may further include performing a
chromatography column run, wherein the raw chromatography data may
be received from the chromatography column run. In other
embodiments, the raw chromatography data may be received from a
chromatography process skid. In still further embodiments, each
block of the plurality of blocks may correspond to a step in a
chromatography process. In further embodiments, the selected
combination may include the first block, the first signal, and a
second signal of the plurality of signals.
In still further embodiments, the method may also include selecting
the combination of the first block and the first signal according
to a profile defining a plurality of selection criteria. In some
embodiments, the plurality of selection criteria may include:
whether blocks occur at regular chromatography cycle intervals; an
extent to which one of the plurality of signals saturates a
detector; an extent to which the plurality of signals approaches a
stationary phase at a distinct level; a magnitude of variation in
the plurality of signals; and/or a number of inflection points
shown by the plurality of signals during a transition phase.
In some embodiments, selecting the combination of the first block
and the first signal may include selecting a primary block and
signal combination, and the method further may include selecting a
secondary block and signal combination. In further embodiments, the
noise reduction technique may include: selecting a portion of the
subset of data to analyze using predetermined set points;
normalizing the portion to prevent magnitude bias; using at least
one smoothing filter on the portion to generate smoothed data; and
analyzing the portion for dynamic signal errors. In yet further
embodiments, the method further may include: selecting smoothed
data matching a feature of a chromatogram transition, wherein the
feature includes one of: derivative duration; maximum intensity;
duration from initiation; or expected background sensor noise. In
still further embodiments, the transition analysis may include
generating a curve using the processed chromatography data, and
analyzing the curve to generate performance parameters.
In some embodiments, the method may further include generating an
Individual chart, a Moving Range chart, or a Range chart based on
the transition data, and generating performance data by applying a
statistical process control to the Individual chart, Moving Range
chart, or Range chart, wherein performing the action based on the
transition data may include performing the action based on the
performance data. In some embodiments, applying a statistical
process control may include performing one of a multivariate data
analysis or a principal component analysis. In some embodiments,
performing an action based on the performance data may include
generating a notification of an event, generating an evaluation of
the event, or generating a deviation notification form. Some
embodiments of the present disclosure may include a chromatography
method that includes performing the process control method while
running a chromatography column.
Some aspects of the present disclosure may relate to a process
control method, the method including: receiving a selection of raw
chromatography data; generating smoothed data by applying a noise
reduction technique to the selection of raw chromatography data,
generating processed chromatography data by selecting smoothed data
matching a feature of a chromatogram transition, and performing an
action based on the processed chromatography data. The noise
reduction technique may include selecting a portion of the smoothed
data to analyze using predetermined set points, normalizing the
portion of data to prevent magnitude bias, using at least one
smoothing filter on the portion of data to generate smoothed data,
and analyzing the portion of data for dynamic signal errors.
In some embodiments, receiving the selection of raw chromatography
data may include receiving raw chromatography data including a
plurality of signals and a plurality of blocks, wherein each signal
of the plurality of signals may be associated with a block, and
selecting a combination of a first block and a first signal from
the raw chromatography data.
In some embodiments, the method further may include using the
processed chromatography data to generate one of an Individual
chart, a Moving Range chart, or a Range chart, and generating
performance data by applying a statistical process control to the
Individual chart, Moving Range chart, or Range chart by performing
a multivariate data analysis or performing a principal component
analysis. In some embodiments, performing the action based on the
processed chromatography data may include performing the action
based on the performance data. In some embodiments, the action may
include generating a notification of an event, generating an
evaluation of the event, or generating a deviation notification
form.
Some aspects of the present disclosure may include a process
control method, the method including receiving processed
chromatography data comprising a combination of a first block and a
first signal, performing a transition analysis on the processed
chromatography data, generating one of an Individual-Moving
Range-Range (I-MR-R) chart based on the transition analysis,
generating performance data by applying a multivariate statistical
analysis method to the I-MR-R chart, and performing an action based
on the performance data. The action may include one of generating a
notification of an event, generating an evaluation of the event, or
generating a deviation notification form.
In some embodiments, the processed chromatography data may comprise
a selection of raw chromatography data to which a noise reduction
technique has been applied. In some embodiments, the selection of
raw chromatography data may be received from a chromatography
process skid.
BRIEF DESCRIPTION OF THE DRAWINGS
The accompanying drawings, which are incorporated in and constitute
a part of this specification, illustrate the disclosed embodiments,
and together with the description, serve to explain the principles
of the disclosed embodiments. In the drawings:
FIG. 1 depicts, in schematic form, an exemplary chromatography
system with which various embodiments of the present disclosure may
be implemented.
FIG. 2 depicts an exemplary chromatogram.
FIG. 3 depicts an exemplary normalized plot of a chromatography
step-up transition.
FIG. 4 depicts a plot of the chromatography step-up transitions of
equilibration conductivity blocks for three lots, according to some
aspects of the present disclosure.
FIG. 5 depicts an exemplary process of analyzing chromatography
data and performing process controls, according to some aspects of
the present disclosure.
FIG. 6 depicts a further exemplary process of analyzing
chromatography data and performing process controls, according to
some aspects of the present disclosure.
FIG. 7 depicts an exemplary data file, according to some aspects of
the present disclosure.
FIG. 8 depicts an exemplary loading plot of a multivariate model,
according to some aspects of the present disclosure.
FIG. 9 depicts an exemplary data smoothing process, according to
some aspects of the present disclosure.
FIG. 10 depicts a loading plot of each variable in a principal
component from 27 lots, according to some aspects of the present
disclosure.
FIG. 11 depicts an exemplary score plot from 27 lots, according to
some aspects of the present disclosure.
FIG. 12 depicts an exemplary loading plot of a multivariate model,
according to some aspects of the present disclosure.
FIG. 13 depicts an exemplary score plot, according to some aspects
of the present disclosure.
FIG. 14 depicts an Individual chart for skewness at a given
chromatography unit operation, according to some aspects of the
present disclosure.
FIG. 15 depicts a Moving Range chart for skewness at a given
chromatography unit operation, according to some aspects of the
present disclosure.
FIG. 16 depicts a Range chart for skewness at a given
chromatography unit operation, according to some aspects of the
present disclosure.
FIG. 17 depicts an Individual chart for non-Gaussian HETP (NG-HETP)
according to some aspects of the present disclosure.
FIG. 18 depicts a Moving Range chart for NG-HETP, according to some
aspects of the present disclosure.
FIG. 19 depicts a Range chart for NG-HETP, according to some
aspects of the present disclosure.
FIG. 20 depicts another Individual chart for NG-HETP, according to
some aspects of the present disclosure.
FIG. 21 depicts yet another Individual chart for NG-HETP, according
to some aspects of the present disclosure.
FIG. 22 depicts an exemplary system on which aspects of the present
disclosure may be implemented.
FIG. 23 depicts an exemplary user interface, according to some
aspects of the present disclosure.
FIG. 24 depicts an exemplary report, according to some aspects of
the present disclosure.
DETAILED DESCRIPTION
The present disclosure relates to improvements in drug product
manufacturing and laboratory processes, as well as improvements in
computer functionality related to drug product manufacturing and
laboratory processes. In particular, aspects of the present
disclosure relate to chromatography methods and systems, and to
methods and systems for chromatography data analysis, e.g., for
monitoring and control of chromatography processes and systems.
Unless otherwise defined, all technical and scientific terms used
herein have the same meaning as is commonly understood by one of
ordinary skill in the art to which this invention belongs. The
materials, methods, and examples are illustrative only and not
intended to be limiting. One of ordinary skill in the art will
appreciate that routine variations on the disclosed materials,
methods, and examples are possible without undue experimentation.
All publications, patent applications, patents, sequences, database
entries, and other references mentioned herein are incorporated by
reference in their entirety. In case of conflict, the present
specification, including definitions, will control.
As used herein, the terms "comprises," "comprising," or any other
variation thereof, are intended to cover a non-exclusive inclusion,
such that a process, method, article, or apparatus that comprises a
list of elements does not include only those elements, but may
include other elements not expressly listed or inherent to such
process, method, article, or apparatus. The term "exemplary" is
used in the sense of "example," rather than "ideal." For such
terms, and for the terms "for example" and "such as," and
grammatical equivalences thereof, the phrase "and without
limitation" is understood to follow unless explicitly stated
otherwise. As used herein, the term "about" and the signifier
".about." are meant to account for variations due to experimental
error. All measurements reported herein are understood to be
modified by the term "about," whether or not the term is explicitly
used, unless explicitly stated otherwise. As used herein, the
singular forms "a," "an," and "the" include plural referents unless
the context clearly dictates otherwise. Moreover, in the claims,
values, limits, and/or other ranges mean the value, limit, and/or
range +/-10%.
As used herein, the term "antibody" includes antigen-binding
molecules as well as antigen-binding fragments of full antibody
molecules. The terms "antigen-binding portion" of an antibody,
"antigen-binding fragment" of an antibody, and the like, as used
herein, include any naturally occurring, enzymatically obtainable,
synthetic, or genetically-engineered polypeptide or glycoprotein
that specifically binds an antigen to form a complex.
Antigen-binding fragments of an antibody may be derived, e.g., from
full antibody molecules using any suitable standard techniques such
as proteolytic digestion or recombinant genetic engineering
techniques involving the manipulation and expression of DNA
encoding antibody variable and optionally constant domains. Such
DNA is known and/or is readily available from, e.g., commercial
sources, DNA libraries (including, e.g., phage-antibody libraries),
or can be synthesized. The DNA may be sequenced and manipulated
chemically or by using molecular biology techniques, for example,
to arrange one or more variable and/or constant domains into a
suitable configuration, or to introduce codons, create cysteine
residues, modify, add or delete amino acids, etc.
Non-limiting examples of antigen-binding fragments include: (i) Fab
fragments; (ii) F(ab')2 fragments; (iii) Fd fragments; (iv) Fv
fragments; (v) single-chain Fv (scFv) molecules; (vi) dAb
fragments; and (vii) minimal recognition units consisting of the
amino acid residues that mimic the hypervariable region of an
antibody (e.g., an isolated complementarity determining region
(CDR) such as a CDR3 peptide), or a constrained FR3-CDR3-FR4
peptide. Other engineered molecules, such as domain specific
antibodies, single domain antibodies, domain-deleted antibodies,
chimeric antibodies, CDR-grafted antibodies, diabodies, triabodies,
tetrabodies, minibodies, nanobodies (e.g. monovalent nanobodies,
bivalent nanobodies, etc.), small modular immunopharmaceuticals
(SMIPs), and shark variable IgNAR domains, also are encompassed
within the expression "antigen-binding fragment," as used
herein.
As used herein, the term "biologic" may refer to a large molecule
(e.g., having a size greater than 30 kDa) created in a living
system such as a cell. Biologics may include proteins (e.g.,
antibodies), nucleic acids, large sugars, etc. Unlike small
molecules that may have well-defined chemical structures, biologics
may have highly complex structures that cannot be easily quantified
by laboratory methods. Thus, it may be desirable to achieve purity,
consistency, and quality in the manufacturing of biologics to
ensure biologic quality, especially when intended for medical
use.
As used herein, the term "chromatography" may refer to any
preparatory or analytical chromatography method. While much of the
present disclosure is provided in the context of preparatory
packed-bed chromatography for purification of a biologic, it is
contemplated that the systems and methods disclosed herein may
apply to a wide variety of chromatography processes.
As used herein, the term "drug product" may refer to a volume of a
formulated drug substance apportioned into a primary packaging
component for packaging, transportation, delivery, and/or
administration to a patient. Drug products may include active
ingredients, including, e.g., biologics.
As used herein, the term "raw material(s)" may refer to a mixture
including one or more biologics, suitable for separation or
purification via a chromatography process.
As used herein, the term "raw chromatography data" may refer to
chromatography data in its native data state as initially
collected. For example, raw chromatography data may be in a .RES
file type, other type of raw file type, or in a database containing
values obtained directly from measurement equipment.
As used herein, the term "extracted chromatography data" can refer
to chromatography data that has been moved from the raw data
without any translation. This can be in an Excel or .CSV file
format, or in a database located within a chromatography system or
computer system.
As used herein, the term "noise reduced data" can refer to
chromatography data, such as transition data, that has been
normalized, smoothed, derived, and/or peak selected.
As discussed above, there exists a need to monitor and maintain
chromatography column and process quality, e.g., over multiple
chromatography runs, over multiple lots, and as time passes both
during and between runs. Systems and methods disclosed herein may
allow for analysis of chromatography transition data (also known as
"transition analysis"), and use of such analyses in monitoring
chromatographic performance, identifying changes in chromatographic
performance, and performing actions with respect to a
chromatography system based on such analyses and processes.
Moreover, systems and methods disclosed herein may, in some
aspects, be a part of one or more in-process manufacturing or
purification controls, and/or may allow for in-process controls
using data collected in standard chromatography processes, thus
minimizing increases in cost and work required to implement
separate process controls.
Reference will now be made in detail to the exemplary embodiments
of the present disclosure described below and illustrated in the
accompanying drawings. Wherever possible, the same reference
numbers will be used throughout the drawings to refer to same or
like parts.
FIG. 1 depicts, in schematic form, an exemplary chromatography
system 100 with which various embodiments of the present disclosure
may be implemented. System 100 includes a mobile phase liquid
supply system 102, a material injection system 104, a column 106, a
process controller 108, a computing device 110, and a detector
112.
System 100 may be all or part of a chromatography system, including
a chromatography column 106. In some instances, system 100 may be a
chromatography skid. System 100 may include any hardware and/or
software required to run a chromatography column. System 100 may be
configured to perform any one of various types of chromatography,
such as high performance liquid chromatography (HPLC), ion exchange
chromatography, size exclusion chromatography, hydrophobic
interaction chromatography (HIC), reverse phase chromatography,
mixed-mode chromatography, or affinity chromatography. System 100
may be used, for example, to separate biologics in a raw mixture,
isolate and/or purify a single type of biologic, and/or eliminate
contaminants from a mixture. In some instances, system 100 may be a
part of a drug product manufacturing system, such as a system for
manufacturing a drug product containing a biologic, such as an
antibody.
Mobile phase liquid supply system 102 may be any suitable system
for supplying a mobile phase to an inlet of column 106. Mobile
phase liquid supply system 102 may include one or more reservoirs
to hold mobile phase liquid(s) used to drive raw materials injected
by material injection system 104 through column 106. Mobile phase
liquid system 102 may include one or more pumps configured to
impart pressure to the mobile phase liquid(s). In some embodiments,
pumps of mobile phase liquid supply system 102 may be configured to
mix two or more solvents (e.g., from two or more reservoirs) in a
desired ratio prior to supplying the combined solution to the inlet
of column 106. In some embodiments, mobile phase liquid supply
system 102 may be configured to supply a first mobile phase to an
inlet of column 106, and then supply a second mobile phase to an
inlet of column 106 after a desired volume of the first mobile
phase has been supplied. In some embodiments, mobile phase liquid
supply system may be controlled by a process controller 108, or by
human interaction.
Material injection system 104 may be any suitable system for
supplying raw material requiring separation and/or purification in
column 106. In some embodiments, for example, material injection
system 104 may include one or more reservoirs to hold raw
materials. Such raw materials may include one or more biologics,
contaminants, solvents, or other materials.
Column 106 may be any column suitable for separating and purifying
injected raw materials from material injection system 104. One of
ordinary skill in the art will recognize that column 106 may have
any of a wide variety of sizes (e.g., diameters ranging from about
30 cm to about 1500 cm) and may be packed with any of a wide
variety of stationary phases. The size, shape, and pack of column
106 may be chosen in view of the raw material requiring separation
in column 106.
Process controller 108 and/or computing device 110 may be suitable
for controlling aspects of system 100 during a chromatography run.
Process controller 108 may be linked to one or more parts of system
100, including mobile phase liquid supply system 102, material
injection system 104, column 106, computing device 110, and
detector 112. In some embodiments, process controller 108 may be a
computer programmed to control parts of system 100 according to a
desired procedure. For example, in some embodiments, process
controller may be programmed to switch pumps of mobile phase liquid
supply system 102 on and off, and to turn detector 112 on and off.
In some embodiments, process controller 108 may have a display
and/or other user interface elements (e.g., buttons, a mouse, a
keyboard, a touch screen, etc.), through which commands may be
input by, e.g., a human operator. In other embodiments, process
controller 108 may be programmed using, e.g., computing device
110.
Computing device 110 may be any computer, such as a desktop
computer, a server computer, a laptop, a tablet, or a personal
portable device (e.g., a smart phone). In some embodiments,
computing device 110 may have a display and/or other user interface
elements (e.g., buttons, a mouse, a keyboard, a touch screen, etc.)
through which commands may be input by, e.g., an operator.
Computing device 110 may also collect data from process controller
108 and/or other parts of system 100, such as detector 112.
Computing device 110 may include one or more programs configured to
display or output such data, e.g., to a screen, a hard disk, or via
an internet connection to a remote location. Computing device 110
itself may be connected to other aspects of system 100 via a wired
connection, or may be wirelessly connected to other aspects of
system 100 (e.g., process controller 108). In some embodiments,
computing device 110 may be located remotely in relation to system
100. In some embodiments, computing device 110 may be configured to
display one or more user interfaces or reports. In some
embodiments, process controller 108 and computing device 110 may be
a single device.
Detector 112 may be any type of detector suitable for detecting one
or more characteristics at the outlet of column 106. Although a
single detector 112 is depicted in FIG. 1, system 100 may include
more than one such detectors configured to detect a variety of
characteristics at the outlet of column 106. Such characteristics
may include, for example, column exit conductivity, pH, optical
density, and other characteristics. In some embodiments, detector
112 may be, for example, an electrical conductivity detector, an
ultraviolet (UV) detector, a fluorescence detector, a refractive
detector, a pH detector, a pressure gauge, or any other type of
detector.
A chromatography cycle, e.g., run using system 100, may typically
include a sequence of steps. Such steps may include, for example, a
cleaning-in-place step, an equilibrium step, a loading step, a wash
step, an elution step, a strip step, and a regeneration step. A
chromatography cycle may be tracked and/or recorded using data
collected from a detector at the outlet of a chromatography column
(e.g., detector 112 at the outlet of column 106). UV detection, for
example, and a UV chromatogram, may be used to track a
chromatography process through, e.g., wash, elution, collection,
and strip steps. FIG. 2 depicts an exemplary UV chromatogram having
a typical profile for collection of a single protein. As volume of
liquid passes through the column (depicted by the x-axis), the UV
detector detects a fairly steady rise in absorbance with a single
peak near the start of the elution step. Collection may be begun
after the small elution peak, during which absorbance spikes as the
majority of the analyte passes the UV detector.
A chromatography (or chromatographic) transition is the response at
the outlet of a column (e.g., column 106) to a change in step at
the column's inlet (e.g., a change from a wash step to an elution
step, or a change from an elution step to a strip step) as one
mobile phase is replaced with another. Depending on what parameters
are being detected at the outlet of a column (by, e.g., one or more
of detector 112), a transition may be detected as an increase (a
step-up transition) or decrease (step-down transition) in one or
more parameters, followed by a plateau of that parameter after
transition has occurred. For example, FIG. 3 depicts an exemplary
normalized plot of a chromatography step-up transition, divided
into three phases. Prior to the transition, a detector detects a
baseline value of a parameter. During transition, the parameter
"steps up" or increases, and then plateaus after transition. In
some cases, the plateau after a step-up transition is due to
detector saturation. The data derived during transition are
quantitative and sensitive to subtle changes in performance of the
column.
Examples of measurable parameters that may change over a transition
include conductivity, pH, salt concentration, light absorption,
fluorescence after excitation with light of a suitable wavelength,
refractive index, electrochemical response, and data generated by
mass spectrometric analysis. One of ordinary skill in the art will
understand, however, that any other measurable parameters that may
change over a transition may be of use in transition analyses
according to the present disclosure.
To perform a transition analysis to determine quality and/or
integrity of a chromatography column and/or process, chromatography
data may be divided into a plurality of blocks, each block
corresponding to a step in the chromatography process (e.g., a
cleaning-in-place block, an equilibrium block, a loading block, a
wash block, an elution block, a strip block, a regeneration block,
a storage block, etc.). Each block includes a plurality of signals
provided by one or more detectors during the block. To perform a
transition analysis, any number or combination of blocks and
signals can be used, such as between 1 and 8 blocks (e.g., 1 block,
2 blocks, 3 blocks, 4 blocks, or 5 blocks), and between about 1 and
8 signals (e.g., 1 signal, 2 signals, 3 signals, 4 signals, 5
signals, 6 signals, or 7 signals.). More blocks and/or signals may
also be used.
FIG. 4 depicts, an exemplary plot of detected conductivity as a
function of volume, during step-up transitions in the equilibration
blocks for three chromatography runs. Each run included the same
chromatography process on the same raw material in the same column,
including isolation of the same protein, but different lots of raw
materials were used. The first spike (in all three runs) represents
a prime of the system. After the spike occurs, as can be seen, the
three runs exhibit variation in the transition phase. The shortest
dashed line depicts the closest to an ideal transition phase, as
the transition is the most "vertical" (i.e., occurs over the
shortest amount of volume). The longer dashed line shows some
characteristics indicative of column failure, namely an early start
to the transition phase, and a tapered ending. Overall, this
transition occurs over a larger amount of volume. The solid line
shows stronger characteristics of column failure, as the transition
phase begins very early and takes excessive time to reach
saturation. While these differences are visually apparent, they may
not be easily quantifiable or given context without being
comparable to one another. The present disclosure provides systems
and methods for performing analyses using these data, and for
reliably performing process controls using such analyses.
FIGS. 5 and 6 depict exemplary processes of analyzing
chromatography data and using such analyses to perform process
controls according to some aspects of the present disclosure. FIG.
5 depicts an exemplary process at a more general level of detail,
whereas FIG. 6 depicts more details of an exemplary process. While
they are described separately below, details and specifics of the
process of FIG. 6 may be applicable to the process of FIG. 5, and
vice versa.
FIG. 5 depicts an exemplary general process 500 of analyzing
chromatography data and performing process controls according to
some aspects of the present disclosure. According to step 510, raw
chromatography data may be processed. According to step 520, data
may be acquired from the raw chromatography data. According to step
530, the acquired data may be processed. According to step 540, the
processed data may be analyzed (e.g., a transition analysis).
According to step 550, one or more statistical process controls may
be performed. According to step 560, data may be reported.
According to step 510, raw chromatography data may be processed.
Raw chromatography data may be obtained by running one or more
chromatography cycles and obtaining signals from one or more
detectors (e.g., detector 112 of column 106). The signals may
comprise, for example, a UV signal, a conductivity signal, a
pressure signal, a pH signal, and/or other signals. The data may be
obtained at, e.g., process controller 108 and/or computing device
110, and may be stored in, e.g., a database or a .RES file. The
data may include, for example, a series of signal values, and
corresponding volumes at which the signal values were measured. The
data may also include indicators of the beginnings and ends of each
block/step in the chromatography cycle.
Processing the data may include extracting the data and organizing
the data in a data file in a computing device, e.g., computing
device 110. Exemplary data files include, e.g., a spreadsheet, a
text file, a database, combinations thereof, and the like. Data
files containing extracted chromatography data may be assigned
various metadata, to allow for consistent storage and processing.
Metadata may include, for example, names, names, dates, column run
times, column run volumes, column mobile phases, identification of
raw mixtures, identification of manufacturing processes for which
the column is used, and/or any other data that may allow for the
consistent automated or manual processing of the data files.
According to step 520, data may be acquired for analysis from the
data files. In some embodiments, an automated software program
(such as Cron, Jobber, a macro, or other automated or scheduling
software) may monitor one or more possible data file storage
locations for one or more data files that fit one or more profiles.
Data files may be assigned a profile based on, e.g., the metadata
associated with the data files. A profile may be, for example, a
pre-made series of selection criteria for selecting one or more
block-and-signal combinations suitable for performing a transition
analysis. A profile may be assigned based on, for example, the type
of column being run, characteristics of the mobile phase, a volume
of mobile phase being run, a column run time, or any other
characteristics of the data files.
Acquiring data for analysis may include selecting one or more
block-and-signal combinations based on an assigned profile, where a
block corresponds to a step in a chromatography process, and a
signal corresponds to a type of data being collected (e.g., UV
data, conductivity, pH, etc.). In some embodiments, a primary
block-and-signal combination may be selected. In further
embodiments, a primary block-and-signal combination and one or more
secondary block-and-signal combinations may be selected. Transition
analysis may be performed first with the primary block-and-signal
combination, and optionally second with the one or more secondary
block-and-signal combinations. Profiles, selection criteria, and
block-and-signal combinations are described in further detail with
respect to process 600.
According to step 530, the acquired data may once again be
processed to obtain noise reduced data. Processing the acquired
data may include applying one or more smoothing and/or noise
reduction techniques to a data set in the acquired data, such as
the data associated with the primary block-and-signal combination,
and optionally the data associated with the secondary
block-and-signal combination. In some embodiments, processing the
data may include standardizing a size of the data set, to allow for
consistent impact of smoothing windows. In some embodiments,
processing the data may include normalizing the data, in order to
eliminate variation based on the magnitude of transitions. This
variation may be due to unique preparations of mobile phase buffers
that contain inherent variability in final value for the baseline
phase or the saturation phase.
Noise reduction techniques may include removal of implicit errors
introduced by measurement tools (e.g., detector 112 in system 100),
and random errors introduced by batch processes when data are
gathered (e.g., in earlier steps of method 500). Noise reduction
may include de-duplication of records in a data set, outlier
detection and removal, and/or any other technique to increase a
signal-to-noise ratio within a data set. Noise reduction may also
include data smoothing and signal rejection, which is described in
further detail below with respect to process 600.
The processed data may include, for example, a step yield and/or
measurements of other mobile phase parameters, which can be in the
form of one or more smoothed curves corresponding to one or more
chromatography step transitions. The one or more curves may
represent a normalized solute signal data array.
According to step 540, the noise reduced data may be analyzed. Such
analysis may be a transition analysis. The transition analysis may
include performing one or more mathematical processes on the
processed data. For example, one or more curves may be generated
from the processed data by, for example, taking a first derivative
of the curve, to generate another curve characterized by a peak.
This curve can be analyzed to generate performance parameters such
as, for example, a number of inflection points, a maximum rate of
change, a breakthrough volume, a cumulative error, NG-HETP, curve
asymmetry, and Gaussian HETP. These performance parameters, either
alone or in combination with past data, may aid in determinations
of column integrity.
For example, an increase in a number of inflection points may
indicate that a slight amount of early breakthrough of a transition
solution is occurring, which may be associated with an integrity
breach. A decrease in maximum rate of change over multiple column
uses may indicate that a transition is taking place over a larger
volume, which can be an indication of an integrity breach. A
decrease in breakthrough volume may characterize an integrity
breach as well. An increase in either NG-HETP or Gaussian HETP may
indicate a decrease in column integrity. Other characteristics of a
transition may be generated based on a modification of the data set
variance, skewness, kurtosis, peak asymmetry, breakthrough or
wash-out volume, and total error. Transition analysis is described
in further detail below. Systems and methods of performing
transition analyses are also described in, e.g., Larson et al., Use
of Process Data To Assess Chromatographic Performance in
Production-Scale Protein Purification Columns, Biotechnol. Prog.,
2003, 19, 485-492, which is incorporated by reference herein in its
entirety.
Results of a transition analysis may be stored, e.g., in a memory
element of computing device 110, or in another computing device,
along with other data. For example, all raw data, initial data
sets, smoothed data sets, and transition analysis data may be
stored.
According to step 550, one or more statistical process controls may
be performed using the results of the transition analysis. In some
embodiments, a statistical process control can include performing
techniques in one of several categories, including 1) a
non-conventional control chart analysis (e.g., an Individual chart,
Moving Range chart, and/or Range chart analysis), 2) a multivariate
analysis (MVA), or 3) a combination of a non-conventional control
chart analysis and MVA. These processes may include, for example,
analyzing the results of the transition analysis as a part of a
larger set of data, including transition analysis results from
prior chromatography runs, e.g., runs in the same production cycle,
runs of the same product lot, or runs of the same raw mixture.
These processes are described with further specificity below, with
respect to process 600.
A result of performing one or more statistical process controls may
be referred to as performance evaluation data. Performance
evaluation data can refer to any process data, in, including
transition analysis results, that have meaning when evaluating the
reproducibility and success of the process.
According to step 560, data may be reported. In some embodiments,
one or more reports may be generated. For example, the methods and
systems disclosed can generate a tabular report of any results
analyzed using a given profile. Reports can be generated based on a
desired number of prior chromatography runs, for a specific
timeframe, for specific runs, and/or for specific lots. An example
report is depicted in FIG. 24, and is described in further detail
below.
FIG. 6 depicts, in further detail than FIG. 5, an exemplary process
600 of analyzing chromatography data and performing process
controls according to some aspects of the present disclosure.
According to step 610, raw chromatography data may be received.
According to step 620, the raw chromatography data may be processed
according to a profile. According to step 630, a noise reduction
technique may be applied. According to step 640, a transition
analysis may be performed on the processed chromatography data to
generate transition data representing a column integrity. According
to step 650, at least one of an Individual (I) chart, Moving Range
(MR) chart, or Range (R) chart may be generated based on the
transition data. According to step 660, one or more multivariate
statistical analysis methods may be applied to the at least one I
chart, MR chart, or R chart to generate performance data. According
to step 660, an action may be performed based on the performance
data.
According to step 610, raw chromatography data may be received. As
with process 500, the raw chromatography may be obtained from,
e.g., a chromatography system such as system 100. The raw
chromatography data may comprise a plurality of signals associated
with a plurality of blocks. Receiving the raw chromatography data
may include directly retrieving raw chromatography data from one or
more detectors (e.g., detector 112 of system 100), or from a
computing device (e.g., computing device 110), and/or may include
monitoring a network location for a raw chromatography data file.
The raw chromatography data may, in some embodiments, be processed,
as described above with respect to step 510 in process 500.
An exemplary data file 1000 of extracted chromatography data is
depicted in FIG. 7. Data file 1000 may include, for example, a data
file name, which may aid in identification of the data file by an
automated system. As shown, extracted chromatography data in data
file 1000 may be in spreadsheet form (e.g., Microsoft Excel). Data
file 1000 may include a volumetric measurement in a first column
1002, which may correspond to periodic measurements of a total
volume that has passed through the chromatography system. A second
column 1004 may include signal measurements (e.g., UV,
Conductivity, pH, etc.) corresponding to each of the volumetric
measurements in column 1002. In this case, second column 1004
contains conductivity data as expressed in mS/cm. Other columns may
provide additional data. Here, for example, a third column 1006
includes volumetric measurements corresponding to logbook entries
in a fourth column 1008. This may allow for identification of
various characteristics of the chromatography run, such as
block/step start and end points (CG002_START, CG002_END,
CG003_START), flow rate, and points at which aspects of the
chromatography system were initiated (e.g., Pump 1 may correspond
to a time when a pump, e.g., associated with mobile phase liquid
supply system 102, is activated). One of ordinary skill will
appreciate that many variations on data file 1000 are possible. For
example, although volumetric measurements are shown in data file
1000 as markers of progress in a chromatography run, other
measurements may be used, such as time. Additional columns for
other signal data may be included, and additional logbook data may
be included (e.g., identifying the mobile phase, identifying the
analyte, etc.)
Referring back to FIG. 6, according to step 620, the chromatography
data may be processed according to a profile. As described briefly
with respect to step 520, a profile may be selected for a
chromatography data file according to characteristics of the
chromatography data in the file. For example, profiles may have
previously been created for a given type of chromatography run, a
given chromatography column, and/or a given analyte. Such profiles
may thus be matched with a chromatography data file for the
appropriate run, column, and/or analyte.
In some aspects, a profile can be created by a user. The profile
may be associated with a specific drug or drug product. In one
aspect, the drug is a small molecule. In other aspects, the drug is
a peptide or a polypeptide.
In some aspects, the drug is a vascular endothelial growth factor
(VEGF) derivative. In other aspects, the drug is aflibercept, which
is described in one or more of U.S. Pat. Nos. 7,070,959, 7,303,746,
7,303,747, 7,306,799, 7,374,757, 7,374,758, 7,531,173, 7,608,261,
7,972,598, 8,029,791, 8,092,803, 8,343,737, and 8,647,842, each of
which is incorporated by reference herein in its entirety.
In other aspects, the drug is an antigen-binding molecule. In some
aspects, the antigen-binding molecule is an antibody or
antigen-binding fragment. In some aspects, the drug is alirocumab,
which is described in U.S. Patent Application Publication Nos.
2014/0356371 and 2014/035670, each of which is incorporated by
reference in its entirety. In another aspect, the drug is
sarilumab, which is described in U.S. Patent Application
Publication Nos. 2016/0152717, 2014/0302053, and 2013/0149310, each
of which is incorporated by reference in its entirety. In another
aspect, the drug is dupilumab, which is described in U.S. Patent
Application Publication No. 2014/0356372, which is incorporated by
reference herein in its entirety. In another aspect, the drug is
selected from the group consisting of evolocumab, bevacizumab,
ranibizumab, tocilizumab, certolizumab, etanercept, adalimumab,
abatacept, infliximab, rituximab, anakinra, trastuzumab,
pegfilgrastim, interferon beta-1a, Insulin glargine [rDNA origin]
injection, epoetin alpha, darbepoetin, filigrastim, and
golimumab.
In some embodiments, a profile may be configured to direct a
sentinel software program (e.g., a macro, Jobber, Cron, or other
scheduling software) to periodically scan a designated network
location for chromatography data files. A profile may direct data
acquisition from a file when the file name matches a file name
identifier in the profile.
Once a profile has selected, or has been selected for or matched
with a data file, the data file may be scanned. For example, with
regard to exemplary data file 1000 in FIG. 7, the fourth column
1008, comprising logbook entries, may be scanned for an indication
of block start times, end times, flow rates, and the like. For
example, with regard to data file 1000, the volumetric measurements
corresponding to "CG002_START" and "CG002_END" bracket the
volumetric measurements that correspond to the chromatographic
operation and signal transition of interest. the first column 1002
and second column 1004 may then be used to extract the full data
set of signals and volume measurements for the operation.
Values in a profile may also define one or more selection criteria
for selecting one or more combinations of blocks and/or signals in
a chromatography data file on which to perform a transition
analysis. Thus, profiles may be tools for acquiring preferable
subsets of data from a chromatography data file. Selection criteria
in a profile may be pre-determined from, e.g., empirical
experience, structured optimization, and/or process documentation.
Such selection criteria may enable identification of
block-and-signal combinations that may allow for more precise,
accurate, or otherwise more useful analyses. Such selection
criteria may include, for example, whether transition materials are
readily available. This includes blocks that transition to, or
transition out of product solutions. This allows for additional
column assessments in between manufacturing operations if so
desired. Such selection criteria may also or alternatively include
whether blocks occur at regular cycle intervals. This includes
blocks that are not performed after the conclusion of a final
collection cycle of a manufacturing lot. Such selection criteria
may also or alternatively include whether signals reach detector
saturation before or after transition. Such selection criteria may
also or alternatively include whether signals approach a stationary
phase at a distinct and identifiable level, and do not continually
drift. Such selection criteria may also or alternatively include
whether signals in a given block have a large difference between
minimum and maximum values. Such selection criteria may also or
alternatively include whether signals have many inflection points
during a transition. Fewer inflection points may indicate more
reliable data collection.
In some instances, prior chromatography runs may assist in
identifying suitable selection criteria for selecting
block-and-signal combinations in future chromatography runs. FIG.
8, for example, illustrates a plot of NG-HETP calculations for two
different block-and-signal combinations (an elution step-UV signal
combination, and a re-equilibration step-conductivity signal
combination) over six different chromatography lots (Lots A-F).
Solid bars denoting three standard deviations for each set are
provided as reference. As can be seen from this plot, the NG-HETP
calculations for the elution step-UV signal combination exhibit
much greater variation than those for the re-equilibration
step-conductivity signal combination. It can be seen that both the
scale of the trends and the standard deviations are different. When
monitoring shifts in performance, it may be desirable to have less
variance across lots that are deemed typical. This allows for
increased sensitivity when monitoring shifts in performance. Thus,
selection criteria for chromatography runs of lots similar to Lots
A-F may include a preference for a re-equilibration step and
conductivity signal combination over an elution step and UV signal
combination. One of skill in the art will appreciate that analysis
of prior chromatography runs in similar fashion may reveal other
potential block-and-signal combination selection criteria.
In some embodiments, a profile may include instructions to apply
one or more selection criteria to a data file having relevant
chromatography data. Thus, processing the chromatography data
according to a profile may include identifying and extracting a
preferred (e.g., primary) block-and-signal combination for
transition analysis, and/or one or more additional (e.g.,
secondary) block-and-signal combinations for transition analysis.
In some embodiments, a primary block-and-signal combination will
meet the most selection criteria in a profile out of all possible
block-and-signal combinations in a chromatography data file. In
some embodiments, a secondary block-and-signal combination will
meet the second most selection criteria in a profile out of all
possible block-and-signal combinations in a chromatography data
file. While a primary block-and-signal combination may include data
most likely to provide a valuable transition analysis for assessing
column and process integrity, a secondary block-and-signal
combination can provide a secondary measurement and a cross-check
of column integrity.
In some embodiments, a profile according to step 620 may be a data
file in and of itself, which may contain instructions for
extracting certain data from, or altering, a chromatography data
file with relevant metadata. In some embodiments, such instructions
in a profile may be executable by a computer program.
Referring back to FIG. 6, after chromatography data has been
processed according to a profile, a noise reduction technique may
be applied to the processed data according to step 630. As with
step 530 of process 500, this step may include applying one or more
smoothing and/or noise reduction techniques to the processed data
(e.g., the data associated with selected block-and-signal
combinations). In some embodiments, this step include standardizing
a size of the data set, to allow for consistent impact of smoothing
windows. In some embodiments, this step may include normalizing the
data, in order to eliminate variation based on the magnitude of
transitions. This variation may be due to unique preparations of
mobile phase buffers that contain inherent variability in final
value for the baseline phase or the saturation phase.
Noise reduction techniques may include removal of implicit errors
introduced by measurement tools (e.g., detector 112 in system 100),
and random errors introduced by batch processes when data are
gathered (e.g., in earlier steps of method 500). Noise reduction
may include de-duplication of records in a data set, outlier
detection and removal, and/or any other technique to increase a
signal-to-noise ratio within a data set.
Noise reduction may also or alternatively include application of a
data-smoothing and signal error-rejection algorithm. FIG. 9
depicts, in flow chart form, an exemplary algorithm 900 in this
regard. According to steps 902 and 904 of algorithm 900, the
algorithm may start, and the relevant signal data (e.g., data that
has been processed according to step 620) is retrieved. According
to step 906, the retrieved data may be normalized to remove
magnitude bias.
A multi-level smoothing algorithm 950 may then be applied. This may
include applying one or more initial smoothing filters (steps 908,
910) according to desired smoothing filter setpoints (909, 911).
According to step 912, a derivation may optionally be performed.
One or more additional smoothing filters may then be applied (steps
914, 916) according to additional desired smoothing filter
setpoints (913, 915). The number of smoothing filters (steps 908,
910, 914, 916) that are applied and the number and characteristics
of setpoints 909, 911, 913, 915 may vary depending on, e.g., data
condition, expected outcomes, signal type, and other factors.
Whether or not a derivation is performed on the data may also
depend on these factors.
Process may then continue to a dynamic signal error-rejection
algorithm 980. This algorithm may be configured to remove data from
the retrieved data that is not due to a chromatographic transition.
For example, errors that should be removed in order to allow for
meaningful transition analysis include alarms, machine arrest, skid
sensor malfunctions, or data gaps. This may be achieved by
identifying the features expected of a chromatogram transition,
such as a derivative duration, a maximum intensity, a duration from
initiation, and expected background noise. For example, an initial
point rejection 918 may be made based on an expected transition
location 919, an initial deadband rejection 920 may be made based
on an expected background noise level 921, a derivative height and
width rejection may be made based on expected signal error
characteristics, and a final deadband rejection may be made based
on expected background noise levels 925. Expected transition
features may be generated, for example, based on prior accumulated
transition data. Upon completion of algorithm 900, according to
step 990, the data may be ready to be used in transition
analyses.
While algorithm 900 is one exemplary model of a smoothing and
signal error-rejection algorithm, one of ordinary skill in the art
will recognize that variations upon this algorithm are possible.
For example, only the smoothing algorithm 950 may be performed, or
only the signal error-rejection algorithm 980 may be performed.
Additionally or alternatively, more or fewer smoothing filters may
be applied, and/or more or fewer points may be rejected.
After applying a noise reduction and/or smoothing technique to the
data, the data may include, for example, step yields and
measurements of other mobile phase parameters in the form of a
breakthrough or washout curve corresponding to a step
transition.
Referring back to FIG. 6, according to step 640, a transition
analysis may be performed on the processed chromatography data to
generate transition data representing a column integrity. The
transition analysis may include performing one or more mathematical
processes on the processed data in order to infer dispersion
parameters from a step transition. For example, one or more curves
may be generated from the processed data by, for example, taking a
first derivative of the curve, to generate another curve
characterized by a peak. This curve may be used to generate
performance parameters such as, for example, a number of inflection
points, a maximum rate of change, a breakthrough volume, a
cumulative error, NG-HETP, curve asymmetry, and Gaussian HETP. As
described with respect to step 540, these parameters may be used as
indicators of column integrity, or a lack thereof (e.g., when
checked against transition analysis parameters of prior
representative chromatography data).
For example, an increase in a number of inflection points may
indicate that a slight amount of early breakthrough of a transition
solution is occurring, which may be associated with an integrity
breach. A number of inflection points may be determined from a
number of peaks when plotting the derivative curve against the
totalized volume data.
As another example, a decrease in maximum rate of change over
multiple column uses may indicate that a transition is taking place
over a larger volume, which can be an indication of an integrity
breach. The maximum rate of change is equivalent to the maximum
value of the derivative curve.
As another example, a decrease in breakthrough volume may
characterize an integrity breach as well. Breakthrough volume may
be determined by finding the first volume value at which the signal
as either less than 95% of its highest value (for a high to low
transition) or greater than 5% of its lowest value (for a low to
high transition).
As another example, an increase in either NG-HETP or Gaussian HETP
may indicate a decrease in column integrity. Other characteristics
of a transition may be generated based on a modification of the
data set variance, skewness, kurtosis, peak asymmetry, breakthrough
or wash-out volume, and total error. Systems and methods of
performing transition analyses are also described in, e.g., Larson
et al., Use of Process Data To Assess Chromatographic Performance
in Production-Scale Protein Purification Columns, Biotechnol.
Prog., 2003, 19, 485-492, which is incorporated by reference herein
in its entirety.
Results of a transition analysis may be stored, e.g., in a memory
element of computing device 110, or in another computing device,
along with other data. For example, all raw data, initial data
sets, smoothed data sets, and transition analysis data may be
stored.
Referring back to FIG. 6, according to step 650, at least one of an
Individual (I) chart, Moving Range (MR) chart, or Range (R) chart
may be generated based on the transition data. For simplicity, this
disclosure will refer to them collectively as an I-MR-R chart;
however, "I-MR-R chart" is to be understood to refer to only an I
chart, only an MR chart, only an R chart, or any combination and
number of such charts. An I-MR-R chart constitutes an individual
visualization of transition analysis outputs, and may aid in
interpreting trends in transition analysis data over multiple
column runs or lots in the form of NG-HETP, skewness, kurtosis, or
other parameters. An advantage of I-MR-R charts is that the data
may be quickly viewable, and may be readily interpretable from a
visual standpoint. This makes slight trends or an immediate data
shift recognizable at an early stage.
An I chart, for example, may plot a value for each analyzed lot
(e.g. skewness). An MR chart may plot a value for the difference
between a value of each analyzed lot and the previously analyzed
lot. An R chart may plot a value for the difference between values
within a lot (e.g., skewness for two transition analyses done on
one lot for a primary block-and-signal combination and a secondary
block-and-signal combination). Each chart may include a mean line,
upper control limits (UCL), and lower control limits (LCL), which
can be calculated using available data that has been determined to
be representative of a typical process, and are placed equidistant
from the mean line in each chart.
Some parameters, when plotted on I-MR-R charts, such as NG-HETP and
skewness of transition analyses, may depict significant dynamics
over the lifetime of certain limits. In such cases, using an I-MR-R
chart with control limits estimated using a short-term standard can
result in excessive out-of-trend signals, even after resetting the
control chart upon repacking of a column. One solution to this
issue is the use of a Levey Jennings control chart, which uses long
term standard deviation calculations from "representative" column
lots that account for special variations attributed to the start-up
of a new column pack. Whether data is considered to be
representative may be determined by having no anomalous readings
for various performance evaluation data sets for a lot. These sets
may be used to calculate standard deviation, sometimes with special
attention to the +1-3 standard deviation (SD) lines. Several lots
may be run on a column to determine whether the majority or entire
useful life of the column was "typical." In one aspect, full
modeling of viable column dynamics can be performed for a Levey
Jennings control chart, which results in a regression model that
accounts for the special cause variation of a column repack. A
Levey Jennings control chart requires longer term data, however,
and thus its use will be limited by the rate of data
aggregation.
Additionally, as transition analysis is known to have variation due
to column repacking events, I-MR-R charts may take into account
packing and repacking of a column--for example, a first lot run
after a column is repacked will not have an MR value that is based
on a change from the last lot run before the column was repacked.
In some aspects, control strategies may be configured to only
consider certain violations that exclude known variation due to
repacking events when monitoring for trending excursions.
Generating of I-MR-R charts may be performed by, e.g., an analysis
module in computing device 110, or in another analysis module
elsewhere. Generation of an I-MR-R chart may also be performed in
computing device 110 by, e.g., a control chart module. For example,
FIGS. 14-21 show I-MR-R data for between 21 and 100 chromatography
lots, and are discussed further below.
Referring back to FIG. 6, according to step 660, one or more
multivariate statistical analysis methods may also be applied to
the I-MR-R data. Alternatively, one or more multivariate
statistical analysis methods may be applied to the transition
analysis data. This step may be performed in addition to, or as an
alternative to, step 650, and like generation of charts according
to step 650, takes into account transition analyses of prior
chromatography data. Multivariate statistical analysis takes
multiple variables and simplifies them to component vectors. This
allows for holistic viewing of large sets of data. Advantages
include that multiple subtle changes across multiple performances,
which would not be evident when looking at singular data sets, may
become evident when graphic their component vectors. Fluctuations
in this data can be caused by differences in materials, equipment,
surrounding atmospheric conditions, and the like, and can be small
from the perception of an operator or human observer. Examples of
multivariate statistical analysis methods may include Principal
Component Analysis (PCA), Partial Least Squares (PLS), Orthogonal
Partial Least Squares (OPLS), Multivariate Regression, Canonical
Correlation, Factor Analysis, Cluster Analysis, Graphical
Procedures, and the like. Such multivariate statistical analyses
may be performed using, e.g., specialized computer software.
The general purpose of using multivariate analysis is to transform
large amounts of data into interpretable information. By enabling a
search for correlations and patterns among multidimensional
variables, and extraction of statistically significant values from
large amounts of raw data, multivariate analysis enables
interpretation of, e.g., any significance to variation between
transition analyses of similar lots of chromatography data.
For example, PCA is a multivariate statistical method where a data
set containing many variables (e.g., a transition analysis
containing several parameters) is reduced to a few variables called
Scores (t). For example, a data set containing many variables may
be reduced to a data set where each observation (e.g., each
transition analysis) is represented by two t-Scores. The t-Scores
contain information about the variation of each variable in the
data set and the correlation of each variable to every other
variable in the data set. As such, t-Scores describe the variation
and correlation structure of each observation (e.g., each
transitional analysis) in the data set to each other observation in
the data set. A graphical output of PCA is commonly a PCA plot. The
PCA plot is a plot of one t-Score against another for each
observation. Generally, the PCA plot is a distribution showing how
the variation and correlation structure compare for all of the
observations in the data set. The plot may thus serve to cluster
similar observations together.
As another example, a PLS regression analysis is a technique for
analysis of systems of independent and response variables. PLS is a
predictive technique which can handle many independent variables,
even when the variables display multicollinearity. PLS may also
relate the set of independent variables to a set of multiple
dependent (response) variables. Often, in PLS, one set of latent
variables may be extracted for the set of manifest independent
variables, and another set of latent variables may be extracted for
the set of manifest response (or dependent) variables. This
extraction process may be based on decomposition of a cross product
matrix involving both the independent and response variables. The
scores, or x-values, of the latent independent variables are used
to predict the scores, or y-values, of the latent response
variables. The predicted y-values are then used to predict
additional manifest response variables. The x- and y-scores are
selected such that the relationship of successive pairs of x- and
y-variables is as strong as possible. The advantages of PLS include
an ability to model multiple independent and dependent variables,
an ability to handle multicollinearity among independent variables,
robustness in the face of data noise and (depending on the software
used) missing data, and creating independent latent variables
directly on the bases of cross-products involving response
variable(s), making for stronger predictions.
In some embodiments, a multivariate statistical analysis may be
performed on an I-MR-R chart, in order to determine further
statistical significance of variation shown in an I-MR-R chart.
In addition to the described analyses above, trends in transition
analysis can be created by calculating non stationary ranges that
allow slow variation to stay within control limits while drastic
shifts to column performance may be flagged as potential out of
trends. Basic methods of defining control limits include moving
average, weighted moving average and various degrees of exponential
smoothing. One such method of calculating trend limits that is
known as the Holt Winters method, or triple exponential smoothing
method can be employed to high effectiveness. The Holt Winters
method employs seasonality for prediction of appropriate limits
that are defined as a discrete column packing event for direct
application to chromatography monitoring. Regression modeling
(e.g., used in the Levey Jennings control chart) constitutes an
additional way to establish trending limits. Once sufficient
empirical data has been obtained, regression modeling of column
integrity can be performed with respect to cumulative column pack
use. This may provide accurate, appropriate ranges of column
performance based on historical column performance included in the
model.
Referring back to FIG. 6, according to step 670, an action may be
performed based on the performance data. In some embodiments, this
can be due to having identified transition analysis as an
in-process control (IPC). In general, an action according to step
670 may include generating a report, generating and/or transmitting
an alert to an operator or to a display, e.g., a display of
computing device 110, or terminating a chromatography process. An
action according to step 660 may also include, e.g., storing all of
the data acquired during systems and methods disclosed herein in a
database, for further analysis.
A result of performing multivariate analysis and/or I-MR-R chart
analysis on transition data can be referred to as performance
evaluation data. Performance evaluation data can refer to any
process data, including transition analysis results, that may have
meaning when evaluating the reproducibility and success of a
process (e.g., a chromatography process).
In one aspect, step 670 may include generating one or more reports.
For example, the methods and systems disclosed can generate reports
in tabular format, of any results analyzed using a given profile.
Reports can be generated based on a desired number of previous
lots, for a specific time frame, and/or for specific lots. The data
sets can be fully extractable into multiple formats and can be
input into external applications if further analysis is
desired.
FIG. 24 depicts an exemplary report 2400 according to some aspects
of the present disclosure. The exemplary report 2400 includes a
Report Pivot Table, that includes the results of four
chromatography cycles from one manufacturing lot. Each of the four
cycles is listed by its lot and cycle number, and includes a
listing of the date and time at which it was run. Transition
analysis results are reported in columns, including NG-HETP,
Gaussian HETP, skewness, asymmetry, kurtosis, Non-Gaussian N, and
Gaussian N. A snapshot of the data source is also provided,
indicating the name of the chromatography system from which the
data came, the logbook in which it was recorded, and the blocks for
which data was taken. Below the data for each of the cycles,
trending data for each of the analysis results is reported. It is
to be understood that this report is an exemplary report, and many
variations are possible. For example, a desired number of
chromatography cycles may be listed and/or included in one or more
plots of trending data.
In some aspects, systems and methods disclosed here may be used for
continuous monitoring of column and process integrity. As such, the
systems and methods disclosed herein can analyze data with respect
to a specific column and/or process. In an aspect, one or more
alerts can be generated based on the data analysis. In another
aspect, the chromatography process can be terminated based on the
data analysis. For example, one or more notifications (e.g., a
notification of event, evaluation of event, or deviation
notification form) can be provided to or displayed to an operator
to take corrective action. For example, one or more screen overlays
can be displayed on, e.g., a screen of computing device 110, and/or
a message may be sent to an operator at the time of analysis
completion, advising on whether to continue or stop a
chromatography process.
In an aspect, results from the disclosed methods and systems can be
trended to impart information of the current trends in assessing
column packing quality prior to column use in manufacturing. In
another aspect, results from the disclosed methods and systems can
be used to evaluate column performance in real-time (or offline)
and can confirm that column integrity prior to the next product use
cycle (e.g., if acceptable range and control limits in a trend
chart are established).
In a further aspect, results can be used with statistical
information to predict process outcomes based on process modeling
using multivariate statistical analysis, prior to expensive and
time-consuming investigation and testing.
One evaluation criterion for statistical analyses plots in
particular, for example, may include that, upon generating a score
plot for a data set using PCA, a lot that is beyond a threshold
number of standard deviations from a mean may be identified as a
column integrity issue, and may cause the generation of an alert or
instruction as to lot variation.
One evaluation criterion for I-MR-R charts in particular may
include that any points outside of upper or lower control limits
for one or multiple chart types may be a basis for an alert. Thus,
the action performed according to step 670 may be to issue an
alert, e.g., from computing device 110, if a lot shows points
outside of control limits. Such alerts may include, for example, a
notification of event, an evaluation of event, and/or a deviation
notification form, to be provided to an operator or a database.
In some aspects, systems and methods disclosed herein may be
implemented as a part of an in-process control system, which may
operate within the framework of an organization's quality system to
ensure consistency and adherence to safety requirements. As a part
of such a program, data from the systems and methods disclosed
herein may be used to determine critical process parameters (CPP)
and critical quality attributes (CQA) to be monitored in an
in-process control program. Additionally, as a part of such a
program, signal transition and column integrity shifts may be
detected in real-time or close to real-time (e.g., while, or
concurrently with, the running of a column), allowing preventative
and corrective actions to be taken in response to performance
data.
FIG. 23 depicts an exemplary user interface 2300 according to some
aspects of the present disclosure. User interface 2300 depicts a
transition analysis profile creation/editing screen with which a
user may generate or edit a new transition analysis profile. The
parameters selected during creation of a profile can be used to
adjust a transition analysis, based on the unique characteristics
of a chromatography process and to optimize robustness of the
output for each column and program. Parameters listed in exemplary
user interface 2300 include, for example, a profile name, comments,
historical data and/or test location, a file pattern, a final
value, a key indicator, a hard reset, a window size for a moving
average, values for a first filter (e.g., an SG filter), values for
a second filter, a percentage of Vmax first under which the signal
should be registered as zero, percentage of max width to retain a
peak, a height of the chromatography column, a start date, an end
date, and a database name.
Methods and systems disclosed herein may be used for relatively
continuous monitoring of column integrity. For example, methods and
systems disclosed herein may monitor column integrity without
requiring interruption of regular chromatography processes to
perform diagnostics on a chromatography system. Moreover, methods
and systems disclosed herein can analyze data with respect to a
specific column and a specific process. As discussed, one or more
alerts may be generated based on data analysis over time. In
another aspect, a chromatography process may be terminated based on
the data analysis. For example, one or more notifications can be
displayed to an operator to take corrective action in the event
that column integrity is found to be compromised. For example, one
or more screen overlays can be displayed and a message window can
be displayed to an operator at the time of an analysis completion,
advising on continuing or stopping a chromatography process, or
advising on other actions.
In some aspects, results from disclosed methods and systems can be
trended to impart information on current trends in assessing column
packing quality prior to column use in manufacturing. In other
aspects, results from the disclosed methods and systems can be used
to evaluate column performance in real-time (or offline) and can
confirm that column integrity prior to the next product use cycle
(e.g., if acceptable range and control limits in a trend chart are
established). In some aspects, results can be used with statistical
information to predict process outcomes based on process modeling
using MVA prior to expensive and time-consuming investigation and
testing.
EXAMPLES
Example 1
A primary block-and-signal combination is chosen from affinity
capture chromatography data of a Protein A as follows. The affinity
capture data includes eight blocks and two signals (UV and
conductivity) in each block, for a total of 16 potential
block-and-signal combination choices. A profile is assigned to the
data, containing a series of block-and-signal selection criteria,
which are applied in the following order to choose a primary
block-and-signal combination: By considering the selection criteria
that blocks must occur at regular intervals among manufacturing
batch cycles, two blocks and their respective signals can be
eliminated, leaving 12 potential combination choices. By
considering the selection criteria that the signal must reach UV
absorbance meter saturation, the UV signal for three blocks can be
removed as candidates, leaving nine potential combination choices.
By considering the selection criteria that signals approach a
stationary phase at a distinct and identifiable level, the UV
signal for three blocks can be removed as candidates, leaving six
potential combination choices (all with conductivity as the signal
choice). By considering the selection criteria that signals should
have a large difference between minimum and maximum values at a
given block, conductivity for four blocks can be removed, leaving
two potential combination choices. By considering the selection
criteria that the signals displaying the least number of inflection
points are preferable, conductivity for one block can be removed,
leaving only one block-and-signal combination choice remaining.
The final remaining block and conductivity signal choice is the
primary block-and-signal combination on which transition analysis
may be performed. The last block-and-signal combination to be
eliminated becomes the secondary block-and-signal combination.
Example 2
I-MR-R trending skewness and NG-HETP data was plotted for 100
chromatography lots in a given chromatography "Program B" as
follows. FIGS. 14-16 illustrate the I, MR, and R charts,
respectively, showing skewness. FIGS. 17-19 illustrate the I, MR,
and R charts, respectively, showing NG-HETP. The UCL and LCL
indicate 3 standard deviations, as determined by previously
accepted data. Breaks in the mean, UCL, and LCL lines indicate a
column repacking. Unbroken shifts in these lines indicate a point
where the limits were recalculated.
FIG. 14 illustrates the skewness for all 100 lots produced in
Program B. It can be seen that the first and second column packs
exhibit different behavior during their use. As shown, Pack 1
experiences a shift in limits after the first four lots, and
maintains skewness values between 0.055 and 0.855. Pack 2 is out of
trend, but eventually reaches a stead state at lot number 67. This
may be due to shifting and settling of the new column pack taking
longer than in Pack 1.
FIG. 15 illustrates an MR Chart for skewness for all lots produced
in Program B. Outliers can be observed for Pack 2 indicating large
shifts between lots based on individual values.
FIG. 16 illustrates an R Chart for the skewness for all lots
produced in Program B. Several outliers are noted in Pack 1. This
increased the limits for Pack 2. There are three packs on the chart
and lots are charted sequentially such that Pack 1 is the leftmost
continuous line and Pack 3 is the rightmost continuous line. Note
that trending points are out during the second half of Pack 1. This
may indicate that the column was experiencing variability within
the cycles of the lots.
FIG. 17 illustrates an I Chart for NG-HETP for all lots produced in
Program B. Pack 1 experiences decreasing NG-HETP, indicative of
improving column behavior. Pack 2 experienced continual increases
in NG-HETP which may have correlated with reduced column
efficacy.
FIG. 18 illustrates a Moving Range Chart for the NG-HETP for all
lots produced in Program B. Outliers can be noticed in both Pack 1
and 2. This identified several points that show dramatic shifts
from individual to individual values.
FIG. 19 illustrates an R Chart for the NG-HETP for all lots
produced in Program B. Pack 2 shows consistently elevated range
values which were investigated and determined to have a root cause
of varying flow direction within the third cycle of the lot. This
caused the third cycle to demonstrate a different value than the
other cycles
Example 3
Individual (I) charts were plotted for transition analyses of two
groups of chromatography lots for a given "Program A."
FIG. 20 illustrates an I chart for the NG-HETP for 46 lots produced
in Program A. The data shows that the column is performed within
established limits for process consistency.
FIG. 21 illustrates an I chart for the NG-HETP for 21 additional
lots produced in Program A. The data shows that two lots (56 and
58) exceeded upper control limits.
Example 4
A multivariate analysis was performed using transition analysis
data from 27 chromatography lots, including the three lots depicted
in FIG. 4. Loading values were calculated for seven parameters from
the 27 lots, including the three lots depicted in FIG. 4. The seven
parameters included NG-HETP for each of an I chart, an MR chart,
and an R chart for the lots, skewness for each of the I chart, MR
chart, and R chart for the lots, and kurtosis for the I chart. FIG.
10 shows a loading chart of each of seven parameters. The magnitude
of each of the bars corresponds to the parameter's effect on the
principal component. Error bars indicate the relative error in the
loading value.
FIG. 11 illustrates an exemplary score plot of the 27 lots. The
score plot was calculated for the seven parameters from 27 lots,
including the loading values calculated for the lots depicted in
FIG. 4 (principal component 1) as well as principal component 2.
Lots with similar parameter values were clustered. The ellipse
around the majority of the plot points excludes outliers with 95%
confidence.
Example 5
A multivariate analysis was performed on I-MR-R data for transition
analysis of 46 chromatography lots as follows. I-MR-R data was
collected for each of the 46 lots. Lots that were deemed atypical
or unsuitable based on the I-MR-R data were removed from the
analysis and data for the remaining lots were collected into Table
1 below. Lots containing values for multiple transitions were
averaged and reported as individual measurements. Range values were
calculated as the maximum values minus minimum values of
transitions within a lot.
TABLE-US-00001 TABLE 1 Individual Individual NG-HETP Skewness Lot
ID NG-HETP Skewness Range Range 1 0.0824 0.73 0.01 0.09 2 0.0951667
0.84 0.007 0.07 3 0.0994 0.826 0.007 0.07 4 0.206167 -0.263333
0.204 1.62 5 0.96625 0.135 2.037 1.75 6 1.91875 -0.10625 3.622 1.83
7 0.107 0.925 0.015 0.06 8 0.55075 0.25625 3.355 2.16 9 0.738625
-0.06375 3.418 2.32 10 0.565714 -0.05 1.302 2.72 11 0.107667 0.715
0.012 0.14 12 0.0745714 0.595714 0.009 0.11 13 0.0651429 0.56 0.006
0.07 14 0.0575714 0.472857 0.002 0.07 15 0.0575714 0.5 0.007 0.17
16 0.054 0.395714 0.008 0.16 17 0.0701429 0.302857 0.115 0.77 18
0.0628571 0.545714 0.019 0.26 19 0.0702857 0.452857 0.018 0.47 20
0.0671429 0.491429 0.033 0.45 21 0.111429 0.172857 0.06 1.19 22
0.167429 -0.101429 0.163 1.41 27 0.1274 0.544 0.009 0.08 28
0.131833 0.575 0.009 0.06 29 0.136667 0.605 0.009 0.05 31 0.133833
0.6 0.011 0.04 32 0.134833 0.595 0.005 0.03 33 0.1368 0.626 0.003
0.04 34 0.135 0.613333 0.003 0.06 35 0.1344 0.632 0.005 0.05 41
0.137667 0.638333 0.014 0.07 42 0.134833 0.62 0.007 0.05 43 0.1352
0.642 0.004 0.03 44 0.1316 0.638 0.008 0.07 45 0.131833 0.641667
0.003 0.07 46 0.135167 0.675 0.009 0.03
Using the data from Table 1, a principal component was calculated
by creating loading plots showing coefficients for each input
parameter. Each row of data was transformed to a single value.
Assessment of model accuracy and relevancy to the physical system
was indicated by R.sup.2 and Q.sup.2 values of the PCA model, where
R.sup.2 is a statistical measure of how close a test set of data
are to the fitted regression line, and Q.sup.2 is a statistical
measure of how close a test set of data would be to the regression
line. Together, R.sup.2 and Q.sup.2 indicate how well a model
describes the system being analyzed, with 1 being perfect modeling
and 0 representing a complete lack of correlation.
FIG. 12 shows a loading plot of the model. The R.sup.2 value for
the model was 0.798, and the Q.sup.2 value was 0.591, indicating
the model was acceptable for use and that all input values had
effects on the model principal component, because they are not
located near the center line. In FIG. 12, the magnitude of the
y-coordinate of each point corresponds to a parameter's effect
(e.g., the effect of average NG-HETP, range of NG-HETP, skewness
range, and average skewness) on the principal component. The y
coordinate of each point corresponds to the number of inputs per
point.
Principal component values were trended and graphed linearly with
respect to corresponding lots. FIG. 13 depicts a score plot for the
data set. The score plot shows the PC1 value (the value contributed
to the direction of highest variance) for each lot used. It can be
seen in FIG. 13 that one lot (Lot 6) was outside a three-standard
deviation limit, and that several points were close to exceeding
two standard deviations, indicating that the system was
experiencing variation in those lots.
As will be appreciated by one of ordinary skill in the art, the
methods and systems disclosed herein may take the form of entirely
hardware embodiments, entirely software embodiments, or embodiments
combining software and hardware aspects. Furthermore, systems and
methods according to the present disclosure may take the form of
computer program products on a computer-readable storage medium
having computer-readable instructions (e.g., computer software)
embodied in the storage medium. Suitable computer-readable storage
media may include hard disks, CD-ROMs, optical storage devices, or
magnetic storage devices. More particularly, the present methods
and systems may take the form of web-implemented computer
software.
Embodiments of the present disclosure are described with reference
to block diagrams and flowchart illustrations of methods, systems,
apparatuses, and computer program products. It will be understood
that one or more blocks of the block diagrams and flowchart
illustrations, respectively, can be implanted by computer program
instructions. These computer program instructions may be loaded
onto a general purpose computer, special purpose computer, or other
programmable data processing apparatus to produce a machine, such
that the instructions which execute on the computer or other
programmable data processing apparatus create a means for
implementing the functions specified in the flowchart block or
blocks.
These computer program instructions may also be stored in a
computer-readable memory that can direct a computer or other
programmable data processing apparatus to function in a particular
manner, such that the instructions stored in the computer-readable
memory produce an article of manufacture including
computer-readable instructions for implementing the function
specified in the flowchart block or blocks. The computer program
instructions may also be loaded onto a computer or other
programmable data processing apparatus to cause a series of
operational steps to be performed on the computer or other
programmable apparatus to produce a computer-implemented process
such that the instructions that execute on the computer or other
programmable apparatus provide steps for implementing the functions
specified in the flowchart block or blocks.
Accordingly, blocks of the block diagrams and flowchart
illustrations support combinations of means for performing the
specified functions, combinations of steps for performing the
specified functions and program instructions for performing the
specified functions. It will also be understood that each block of
the block diagrams and flowchart illustrations, and combinations of
blocks in the block diagrams and flowchart illustrations, can be
implemented by hardware-based computer systems that perform the
specified functions or steps, or combinations of hardware (e.g.,
special-purpose chromatography hardware) and computer
instructions.
FIG. 22 depicts an operating environment 2200 in which some systems
and methods according to the present disclosure may be implemented.
By way of example, process controller 108 and computer device 110
(or a component thereof) of FIG. 1 could be a computer 2201, as
illustrated in FIG. 22. Computer 2201 can comprise one or more
components, such as one or more processors 2203, a system memory
2212, and a bus 2213 that couples various components of a computer
2201 including the one or more processors 2203 to the system memory
2212. In the case of multiple processor 2203, the system can use
parallel computing.
The bus 2213 can comprise one or more of several possible types of
bus structures, such as a memory bus, memory controller, a
peripheral bus, an accelerated graphics port, and a processor or
local bus using any of a variety of bus architectures. The bus
2213, and all buses specified in this description can also be
implemented over a wired or wireless network connection.
The computer 2201 typically comprises a variety of computer
readable media. Exemplary readable media can be any available media
that is accessible by the computer 2201 and comprises, for example
and not meant to be limiting, both volatile and non-volatile media,
removable and non-removable media. The system memory 2212 can
comprise computer readable media in the form of volatile memory,
such as random access memory (RAM), and/or non-volatile memory,
such as read only memory (ROM). The system memory 2212 typically
can comprise data such as chromatography data 2207 and/or program
modules such as operating system 2205 and chromatography software
2206 that are accessible to and/or are operated on by the one or
more processors 2203. The many features and advantages of the
present disclosure are apparent from the detailed specification,
and thus, it is intended by the appended claims to cover all such
features and advantages of the present disclosure that fall within
the true spirit and scope of the disclosure. Further, since
numerous modifications and variations will readily occur to those
skilled in the art, it is not desired to limit the present
disclosure to the exact construction and operation illustrated and
described, and accordingly, all suitable modifications and
equivalents may be resorted to, falling within the scope of the
present disclosure.
In another aspect, the computer 2201 can also comprise other
removable/non-removable, volatile/non-volatile computer storage
media. The mass storage device 2204 can provide non-volatile
storage of computer code, computer readable instructions, data
structures, program modules, and other data for the computer 2201.
For example, a mass storage device 2204 can be a hard disk, a
removable magnetic disk, a removable optical disk, magnetic
cassettes or other magnetic storage devices, flash memory cards,
CD-ROM, digital versatile disks (DVD) or other optical storage,
random access memories (RAM), read only memories (ROM),
electrically erasable programmable read-only memory (EEPROM), and
the like.
Those skilled in the art will appreciate that the conception upon
which this disclosure is based may readily be used as a basis for
designing other structures, methods, and systems for carrying out
the several purposes of the present disclosure. Accordingly, the
claims are not to be considered as limited by the foregoing
description.
* * * * *